Several new features are added to the long term High Speed Packet Access (HSPA) evolution in order to meet performance requirements set by the International Mobile Telecommunications Advanced (IMT-A). The main objective of these additional features is to increase average spectral efficiency. One possible technique for improving downlink spectral efficiency is to introduce support for four branch multiple-input multiple-output (MIMO), i.e. utilize up to four transmit and receive antennas to enhance the spatial multiplexing gains and to offer improved beamforming capabilities. Four branch MIMO provides up to 84 Mbps per 5 MHz carrier for high signal to noise ratio (SNR) users and improves the coverage for low SNR users.
The current HSPA system (Release 7-10) supports 1 or 2 transmit antennas at the base station (e.g., NodeB). For these systems, from channel sounding, a user equipment (UE) (e.g., a mobile phone or any other wireless communication device) measures a channel and reports in one sub frame channel state information related to the measured channel. Typically, this report consists of a channel quality indicator (CQI), which explicitly indicates rank indicator (RI), and precoding control indicator (PCI). The UE sends this report periodically for every subframe (TTI). The NodeB uses this report in scheduling the UE.
Introduction of 4-branch MIMO will require a new feedback channel structure to send the CQI/PCI information to the NodeB. To reduce the signaling overhead at downlink and uplink, it was recommended to use two codewords for four branch MIMO. For the uplink signalling channel (HS-DPCCH), it was agreed to use a structure similar to that of 2 Antenna MIMO (Release-7), where the CQI, PCI are sent in one subframe. Hence in total 14 bits are used to report: CQI (8 bits), RI (2 bits) and PCI (4 bits). Since the total number of bits is different compared to that of Release 7 structure of 10 bits, a new encoding is scheme is needed to transmit these bits. Hence to encode these 14 bits in one subframe a convolutional code with 1/3 code rate has been proposed. Unfortunately the puncturing pattern chosen for this scheme is not optimal in terms of packet error performance.